FR3052007A1 - METHOD AND DEVICE FOR RECEIVING AUDIOVISUAL CONTENT AND CORRESPONDING COMPUTER PROGRAM - Google Patents

Abstract

The invention relates to a method for receiving audiovisual content from an audiovisual stream that includes a video component and an audio component, a subtitle component being generated in association with the audio component and containing a plurality of words representative of audio data of the audio component, said method being characterized in that it implements a restitution of the content, during which between two successive instants of restitution of a current video scene of the video component, in the case where the common audio data delivered between said two successive instants are representative of a plurality of K words, such that K≥2, said K words are restored between said two successive instants, respectively at K successive instants, a word restored at a given instant replacing the word restored at a time immediately preceding said given instant.

Description

Method and device for receiving audiovisual content and corresponding computer program

Field of Invention

The field of the invention is that of the reproduction of audiovisual contents by a terminal, from an audiovisual stream which comprises a video component and an audio component, a subtitling component being generated in association with the audio component and containing a series of words representative of the audio data of the audio component. The invention applies both to audiovisual content broadcast in real time on a user terminal, as audiovisual content previously recorded in the latter.

More particularly, the invention applies to the way in which the words of the subtitling component are rendered during the rendering of the content. The invention can in particular be implemented in a terminal with a user interface and a graphical interface, for example a tablet, a mobile phone, a smartphone ("smart phone"), a personal computer, a connected television to a communication network, etc.

Presentation of the prior art

Today, the techniques of subtitling audiovisual content allow a user, at a current moment of rendering said content, to display on a display screen, along with the content, one or more sentences of retranscription of the dialogue pronounced by the character or characters intervening in the content, at the current moment. Such retranscription sentences are displayed on one or more lines, depending on their length, and are generally arranged at the bottom of the display screen.

A disadvantage of such a method lies in the fact that the viewing comfort of the user is not satisfactory. Indeed, the user spends his time doing eye movements from left to right, which is a physical constraint that disrupts the monitoring of audiovisual content, for example a film, and causes additional eye strain. The display of subtitles in the form of sentences also imposes a maximum size of the font, which can be difficult to read by users with visual problems.

Another disadvantage of the current captioning techniques is that the user reads at his own pace the subtitle sentence or sentences that appear on the screen, which can cause a desynchronization and a lack of rhythm between the sentences that are read and the words actually spoken.

On the other hand, when a sentence pronounced by a speaker does not have a constant bit rate, the speaker being able for example to pronounce the beginning of the sentence very quickly and the end of the sentence more slowly, the user will not be able to distinguish this break. rhythm by reading the subtitled sentence.

In addition, certain dialogues are not systematically transcribed accurately, and are sometimes transcribed or translated with shorter sentences to respect the space available for subtitles.

All these disadvantages therefore do not allow the user to experience a real feeling of immersion in relation to the content that is returned to him because the current subtitling in the form of whole sentences tends to distort the initial content.

Object and summary of the invention

One of the aims of the invention is to overcome disadvantages of the state of the art mentioned above. For this purpose, an object of the present invention relates to a method for receiving audiovisual content from an audiovisual stream which comprises a video component and an audio component, a subtitling component being generated in association with the audio component and containing a series of words representative of the audio data of the audio component.

Such a method is remarkable in that it implements a restitution of the content, during which between two successive instants of restitution of a current video scene of the video component, in the case where the current audio data delivered between the two successive instants are representative of a plurality of K words, such that K> 2, the K words are restored between the two successive instants, respectively at K successive instants, a word restored at a given instant replacing the word restored at a given instant immediately preceding said given instant.

Such a disposition advantageously makes it possible, as and when the retrieval of a given content, to reproduce textually on the screen each spoken word, one by one, a word being restored in place of the word which has been restored just before . The viewing comfort of the user is significantly improved since he no longer has to turn his eyes constantly from left to right to read the subtitle sentences.

In addition, such an arrangement allows a dynamic adaptation of the rate of display of words to the actual flow of the word or words uttered by the speakers involved in the content.

Thus the user can also become aware of the speed of speech of each speaker, whites, hesitations in pronounced sentences, which will allow him to better feel the character / behavior of each character, the actual lived scenes, etc. In particular, the sensation of immersion of the user with respect to the content is much more marked in comparison with prior art content rendering methods.

Restitution means viewing a content, listening to content, or both at the same time.

A video scene is a sequence of several images that scroll over a given time range.

The reception method according to the invention is for example implemented in a terminal, such as a set-top-box or in a terminal connected to the set-top box, such as for example a tablet, a television set. , etc ....

According to a particular embodiment, the subtitling component being generated prior to the return of the content, each of the K words, pronounced at successive K instants, having previously been associated with an identifier representative of the speaker who pronounces the word, the restitution content at a moment considered among the K instants, is implemented: - by a prior analysis of the identifier corresponding to the instant considered, - by triggering a display command, at the moment considered, the word associated with the analyzed identifier, in visual correspondence with the speaker associated with the analyzed identifier.

Such a provision advantageously allows the user to easily distinguish on the screen which speaker speaks the word displayed at the moment considered, especially when several speakers speak almost simultaneously or several speakers appear at the same time to the screen.

According to yet another particular embodiment, the subtitling component being generated prior to the return of the content and the plurality of K words being pronounced by the same speaker at successive K instants, the first word of the plurality of K words having in addition, previously associated with an identifier representative of the same speaker, the restitution of the content is implemented: - at the first moment among the K instants, by a prior analysis of the identifier corresponding to the first word of the plurality of K words by triggering a command for successively displaying K words, respectively at K instants, in visual correspondence with said same speaker associated with the analyzed identifier.

Such a provision advantageously allows the user to easily distinguish on the screen which speaker speaks the word displayed at the current time, in the case where several people are displayed on the screen.

Furthermore, the fact of not systematically associating a speaker identifier with each word uttered subsequently by the same speaker makes it possible to greatly reduce the cost of signaling the audiovisual stream.

According to yet another particular embodiment, the subtitling component being generated prior to the return of the content and the plurality of K words consisting of: - a first plurality of J words uttered respectively J moments by a first speaker at least a second plurality of L words uttered respectively at L times by a second speaker, following the J words uttered by the first speaker, the first word of the plurality of J words having been previously associated with a first speaker; representative identifier of the first speaker and the first word of the plurality of L words having previously been associated with a second identifier representative of the second speaker, the restitution of said content is implemented: - at the first instant of the plurality of J instants, by a preliminary analysis of the first identifier associated with the first word of the plurality of J words, - by triggering a comma n of successive display of the J words, respectively J instant, in visual correspondence with the first speaker associated with the first identifier analyzed, - at the first moment of the plurality of L instants, by a prior analysis of the second identifier associated with the first word of the plurality of L words, - by triggering a command for successive display of L words, respectively at L instants, in visual correspondence with the second speaker associated with the second identifier analyzed.

Such an arrangement allows the user, when at least two speakers speak successively, to identify the change of speaker clearly and easily.

Furthermore, the fact of not systematically associating a speaker identifier with each word uttered subsequently by the same speaker makes it possible to greatly reduce the cost of signaling the audiovisual stream.

According to yet another particular embodiment, the subtitling component being generated prior to the return of the content and the plurality of K words being pronounced by the same speaker at K successive successive times, at a given speed, the given speed having in addition, previously associated with an identifier representative of the value of the given speed, the restitution of the content is implemented: - by a preliminary analysis of the identifier representative of the value of the given speed, - by a calculation, in a function of the value of the speed associated with the analyzed identifier, of each of the K instruction instants of the K words, - by triggering a command for successive display of the K words, respectively at the K instants calculated.

Such an arrangement advantageously makes it possible to preserve the structure of the existing subtitling components, that is to say which appear in the form of groups of successive sentences, while restoring each spoken word verbatim on the screen, one by one, as and when restitution of a given content, each word restored at a given instant replacing the word which is restored at a time immediately preceding the moment considered.

Such a feature is made possible by the fact that the sentences of a conventional subtitling component are advantageously previously divided into sub-sentences respectively corresponding to different pronunciation rates.

In addition, each word of a spoken sentence is advantageously displayed at the actual pronunciation rate of this sentence, which significantly improves the user's feeling of immersion with respect to the content.

According to yet another particular embodiment, the subtitling component is generated simultaneously with an analysis of the content, by the implementation of the following: identification, at a given moment among the K K word pronunciation instants : • one or more audio data of the audio component delivered at the moment considered and representative of one of the K words, • of the corresponding video scene synchronized with the audio data or data delivered, - textual conversion of the or audio data delivered at the instant in question, in one word, triggering a synchronized reproduction control of the audio data or data, the corresponding video scene and the word, in association with the instant in question.

Such an arrangement makes it possible to avoid transmitting a subtitling component in the audiovisual stream, which makes it possible to greatly reduce the cost of signaling the audiovisual stream, while at the same time advantageously allowing, at a given moment of restitution of a given content. , to reproduce textually on the screen a word spoken at a time, synchronized with the sound and the image delivered at the instant of restitution considered.

According to yet another particular embodiment, the analysis of the content comprises, at the instant considered among the K instants of pronunciation of the K words: an identification of the voice frequency associated with the corresponding audio data or data delivered and representative of one of the K words uttered by a speaker considered, - a matching of the identified voice frequency with information relating to the speaker in question which has been previously associated with the voice frequency, the reception method implementing, at the moment considered restitution, a triggering of a word display command corresponding to the audio data / delivered, in visual correspondence with the speaker considered.

Such a provision advantageously allows the user, when the subtitling component is generated in real time, to easily distinguish on the screen which speaker speaks the word displayed at the current time, especially when several speakers speak almost at the same time or when several speakers are displayed on the screen.

The aforementioned different embodiments or features may be added independently or in combination with each other to the reception method defined above. The invention also relates to a device for receiving audiovisual content from an audiovisual stream that comprises a video component and an audio component, a subtitling component being generated in association with the audio component and containing a plurality of K words. representative of the audio data of the audio component (K> 2).

Such a reception device is remarkable in that it comprises a processor which is arranged to implement a restitution of the content, during which between two successive instants of restitution of a current video scene of the video component, in the case where the current audio data delivered between the two successive instants are representative of a plurality of K words, such that K> 2, the K words are restored between the two successive instants, respectively at K successive times, a word restored to a given instant replacing the word restored at a time immediately preceding the given moment. The invention also relates to a computer program comprising instructions for implementing the reception method according to the invention, when it is executed on a terminal or more generally on a computer.

Each of these programs can use any programming language, and be in the form of source code, object code, or intermediate code between source code and object code, such as in a partially compiled form, or in any form what other form is desirable. The invention also relates to a recording medium readable by a computer on which a computer program is recorded, this program comprising instructions adapted to the implementation of the reception method according to the invention, as described above. .

Such a recording medium may be any entity or device capable of storing the program. For example, the medium may comprise storage means, such as a ROM, for example a CD ROM or a microelectronic circuit ROM, a USB key, or a magnetic recording means, for example a hard disk. On the other hand, such a recording medium may be a transmissible medium such as an electrical or optical signal, which may be conveyed via an electrical or optical cable, by radio or by other means. The program according to the invention can be downloaded in particular on an Internet type network.

Alternatively, the recording medium may be an integrated circuit in which the program is incorporated, the circuit being adapted to execute or to be used in the execution of the aforementioned reception method.

The receiving device, the computer program and the corresponding recording medium mentioned above have at least the same advantages as those conferred by the reception method according to the present invention.

List of Figures Other advantages and characteristics of the invention will appear more clearly on reading the following description of several particular embodiments of the invention, given as simple illustrative and non-limiting examples, and the accompanying drawings. among which: - Figure 1 shows schematically an architecture in which is implemented the method of receiving audiovisual content according to the invention; FIG. 2 schematically shows the steps of a method for generating audiovisual streams according to a first embodiment of the invention; FIG. 3 represents the structure of the audiovisual stream generated according to the method of FIG. 2; FIGS. 4A to 4C show different variants of the subtitling component of the audiovisual stream generated according to the method of FIG. 2; FIG. 5 schematically shows the steps of a method for generating audiovisual streams according to a second embodiment of the invention; FIG. 6 represents the structure of the audiovisual stream generated according to the method of FIG. 5; FIGS. 7A to 7D represent different variants of the subtitling component of the audiovisual stream generated according to the method of FIG. 5; FIG. 8 schematically shows the steps of a method for generating audiovisual streams according to a third embodiment of the invention; FIG. 9 represents the structure of the audiovisual stream generated according to the method of FIG. 8; FIG. 10 shows the simplified structure of an audiovisual stream receiving device according to one embodiment of the invention; FIG. 11 schematically shows the steps of an audiovisual stream reception method according to a first embodiment of the invention; FIGS. 12A and 12B respectively represent two examples of subtitled video scenes as restored in two different variants of the reception method of FIG. 11; FIG. 13 schematically shows the steps of an audiovisual stream reception method according to a second embodiment of the invention; FIGS. 14A and 14B respectively represent two examples of subtitled video scenes as restored according to two different variants of the reception method of FIG. 13; FIG. 15 schematically shows the steps of an audiovisual stream reception method according to a third embodiment of the invention.

Description of particular embodiments of the invention

In relation with FIG. 1, an architecture is presented in which the method for receiving audiovisual contents according to the invention is implemented.

Such an architecture includes a TER terminal for accessing content offered by a PFS service platform, via an RC communication network, such as for example of the IP (abbreviation of "Internet Protocol") type. The service platform PFS offers different contents to the user UT of the terminal TER, such as for example: television television contents, in particular those for broadcast programmed into a program schedule, such as films, sports events, programs , television newscasts, etc ... - VOD catalogs, - video catalogs, - music catalogs (clips, concerts, etc ...) - podcast catalogs, - digital book catalogs, - catalogs of applications and / or services. The aforementioned architecture allows the user UT terminal TER to have access to content offered both in a situation of mobility in a sedentary situation.

In a situation of mobility, the TER terminal is for example a mobile phone, a smartphone ("smart phone"), a tablet, a laptop, etc.

In a sedentary situation, the TER terminal could be a PC-type personal computer.

Still in a sedentary situation, and as represented in FIG. 1, the TER terminal is composed for example of: an access terminal STB which is able to receive and process the contents coming from the platform PFS, a reproduction terminal, for example a TV TLV as shown in Figure 1, adapted to restore the UT user the contents processed by the access terminal STB.

In an exemplary embodiment, the access terminal and the rendering terminal are grouped into a single terminal. It could, for example, be a television containing a set-top-box decoder. In another example, the access terminal STB is a set-top box and the terminal TER is a tablet. office of restitution terminal connected to the set-top-box by means of a local network, for example wireless, in particular of the WiFi or CPL type (abbreviation of "power lines in line"). According to other examples not shown, the TER terminal could be a mobile phone, a smartphone ("smart phone"), the TV TLV or a radio connected to a communication network, etc ... The UT user can interact with the access terminal STB using a conventional remote control or using the terminal TER which includes for this purpose a suitable software application remote control. The TER terminal then has the possibility of displaying an interface containing keys dedicated to prerecorded commands. Thus, the TER terminal has the same functions as a conventional TV remote control. For example, the user can request the selection of content received from the PFS service platform by simply pressing the "i- >>, >>," i "," i "directional keys in an associated menu. viewing and / or listening to received content. The user can also validate the selected content by pressing the "OK" key. When the user activates a key on his remote control, a message comprising the command associated with this key is sent to the access terminal STB according to a communication protocol adapted to the local network used.

The access terminal STB, as well as the terminal TER, furthermore comprise means of connection to the communication network RC which may be, for example, of the x-DSL, fiber or 3G and 4G type.

With reference to FIG. 2, the steps of the method for generating an audiovisual stream Fi representative of an audiovisual content G adapted to be received by the access terminal STB of the present invention are presented according to a first embodiment. Figure 1, via the communication network RC.

Such audiovisual content Ci of a duration T, consists for example of a film, a television program, the retransmission of a speech, a concert, a documentary, a sports event such as in particular a football match, etc. .....

The method for generating the audiovisual stream Fi comprises an editing step EDi during which the stream Fi is built by inserting: a video component CVi which contains all the video data contributing to the restitution of the images of the content audiovisual G, - at least one audio component CAi which contains all the audio data contributing to the complete sound reproduction of audiovisual content G.

The video components CVi and audio CAi are inserted in a synchronized manner so that at a current moment of restitution of the audiovisual content, the sound emitted corresponds to the image rendered on the screen.

Eraser represented in FIG. 3: the video component GVi contains a sequence of M video scenes SVi, SV2,..., SVu,... SVm which scroll over the entire duration T of the content G, respectively at different successive times ti , ta, ..., you, ..., tM, - the audio component GAi contains a sequence of M sets of audio data Di, D2, ..., Du, ... Dm which follow each other over the duration T of the content G and which are respectively synchronized with the M video scenes of the video component GVi at the different successive instants ti, ta, ..., tu, ..., tM which constitute temporal indicators of synchronization between the data of the GVi video component and the audio data of the audio component GAi.

Each of the M video scenes consists of a sequence of several images scrolling at a given rate (24 frames per second for example). In addition, the number of images per video scene is not necessarily the same from one video scene to another.

Each of said M sets of audio data, or only some of them, contain audio data that is associated with one or more words uttered by a speaker or several speakers involved in the content C. By speaker involved in the content G we mean a person who pronounces the word or words being visible on the corresponding image or a person who pronounces the word or words, without being visible on the corresponding image ("voice off").

During a signaling step SIG1 shown in FIG. 2, it is moreover conventionally performed to insert, in a signaling sub-flow SFi associated with the stream Fi, a plurality of information allowing the analysis of the Fi stream by the TER terminal or the access terminal STB of the user. Such information is in particular metadata, such as, for example, an identifier ID of the audiovisual content C, description information DESC of the audiovisual content C, for example its genre (film, documentary, sport, etc.), the name and the first name of the persons associated with the audiovisual content C (director, actors, actresses, sportsmen, etc.), associated temporal information, such as, for example, the duration T of the audiovisual content C, the start time and end of broadcast of content C if it is broadcast in real time, etc ...

Steps ED1 and SIG1 are known as such.

According to the first embodiment of the invention, the audiovisual stream generation method further comprises a step ST1 of subtitling the audiovisual content C, during which a subtitling component CSi is generated in association with the audio component CAi. In a manner known per se, the subtitling component CSi conveys the text transcript of sentences pronounced over the duration T of audiovisual content C.

According to the first embodiment of the invention, as represented in FIG. 3, at a current instant tu of appearance of a video scene SVu, and until the following moment tu + i of appearance of the scene next video SVu + i, in the case where the corresponding set of audio data of the audio component CAi contains audio data representative of a plurality of K words mki, mk2, ..., mj, ..., mK constituting one or more sentences later, during the subtitling step ST1 of FIG. 2, the K words are indexed temporally one by one, respectively at K successive moments tki, tk2, ..., tK, so as to be restored later on the screen, one by one, between the two successive instants tu and tu + i of appearance of the video scene SVu, a word restored at a given moment among the K instants replacing the word restored to the instant immediately preceding the given instant. Step ST1 is thus repeated for each time interval [t1, ta], [ta, ta], and so on until the last time interval [tM, Tj.

According to a first variant of this first embodiment, which is represented in FIG. 4A, the K words mki, mka, ···, mK are respectively associated, in the subtitling component CSi, with K identifiers lki, Ik2 , ···, Ik, such that: - the identifier lki is representative of the speaker who pronounces the word mki, - the identifier Ika is representative of the speaker who pronounces the word trika,

J - the identifier Ik is representative of the speaker who pronounces the word nriK.

Such an identifier consists for example in: - the name of the speaker, and / or - a particular display color of the word associated with said identifier, and / or - the coordinates of an area of the video scene SVu in which is intended to displayed the word associated with said identifier considered.

In the case where the speaker who pronounces one of the K words, for example the word mki, is visible in the video scene SVu, either alone or with other persons, the coordinates of said display zone of the word mki are determined so that the word mki is displayed on the screen in a video zone that is arranged close to said speaker.

In the case where the speaker who pronounces one of the K words, for example the word mki, is not visible in the video scene SVu ("voice off" for example), the coordinates of said display area of the mki word are determined so that the word mki appears on the screen in a neutral zone of the image, for example at the bottom of the latter.

According to a second variant of this first embodiment which is represented in FIG. 4B, in the case where the K words are all uttered by the same speaker, only the first word mki among said K words is associated with a lki identifier of the speaker who pronounces the K words.

According to a third variant of this first embodiment which is represented in FIG. 4C, in the case where the plurality of K words consists of a first plurality of J words mji, mj2, ..., mj pronounced by a first respectively at successive instants tji, tj2, ..., tj and at least a second plurality of L words mn, mi2, ..., vni pronounced by a second speaker, following the first speaker, respectively at The successive instants tu, ti2, ..., tk, - at least the first word mji among said J words is associated with an identifier lji of the first speaker who pronounces the J words, - at least the first word mn among said L words is associated with an identifier In of the second speaker who pronounces the L words.

In order to predict the situation where a user would access the content in the time interval [tji, tj] (respectively [tu, tJ), but after the instant tji (respectively tii), each of the following words mj2 to mj (respectively mi2 to ϊΏι) is also associated with the identifier lji of the first speaker who pronounces the J words (respectively the identifier In of the second speaker who pronounces the L words). At the end of the subtitling step ST1 of FIG. 2, the stream Fi comprising the video component CVi, the audio component CAi, the subtitling component CSi, in association with the sub-stream SF, is generated. ^. The flow Fl contains the temporal synchronization indicators ti, ta, ..., tu, ..., tM, as well as all the instants of pronunciation of the words.

The stream Fi is then either transmitted to the terminal TER or to the access terminal STB of FIG. 1, via the network RC, or stored on a medium for a subsequent transmission.

It is recalled that the invention can be applied to audiovisual streams transmitted or broadcast live (in English "live") or downloaded streams.

A second embodiment of a method for transmitting audiovisual streams according to the invention will now be described with reference to FIG.

Similarly to the first embodiment shown in FIG. 2, the method of generating audiovisual streams according to the second embodiment comprises an editing step EDio during which a stream F'i is built by insertion of: video component CVio which contains all the video data contributing to the restitution of the audiovisual content images G, at least one audio component CAio which contains all the audio data contributing to the complete sound reproduction of the audiovisual content G.

The video components CVio and audio CAio are inserted in a synchronized manner so that at a current moment of restitution of the audiovisual content, the sound emitted corresponds to the image restored on the screen.

As represented in FIG. 6: the video component CVio contains a sequence of M video scenes SVi, SV2,..., SVu,... SVm which scroll over the duration T of the content G, respectively at different successive instants ti , ta, ..., you, ..., tM, the audio component CA10 contains a sequence of M audio data sets Di, Da, ..., Du, ... Dm which succeed each other over the entire duration T of the content G and which are synchronized respectively with the M video scenes of the video component CVio, at different successive times ti. Ϊ2, ..., tu, ..., tM which constitute temporal indicators of synchronization between the data of the video component CVio and the audio data of the audio component CAiq.

Similarly to the first embodiment shown in FIG. 2, the transmission method according to the second embodiment further comprises a signaling step SIG10 represented in FIG. 5, during which it is conventionally carried out at the insertion, in a signaling sub-flow SF'i associated with the flow F'i, a plurality of information allowing the analysis of the flow F'i by the access terminal STB of the user. Such information is identical to the information reported in the first embodiment.

The stream generation method according to the second embodiment further comprises a step ST10 of subtitling audiovisual content C, during which a caption component CSio is generated in association with the audio component CAio of the same way than in the prior art. For this purpose, as represented in FIG. 6, at a current instant tu of appearance of a video scene SVu, and until the next instant tu + i of appearance of the next video scene SVu + i, in the case where the corresponding set of audio data of the audio component CAio contains audio data representative of one or more words, this or these words are retranscribed textually in the form of one or more sentences, noted PFlu on the figure 6, between the moment you and the moment tu + i. For example, a single sentence PFlu is pronounced between the instant tu and the instant tu + i, said sentence containing K words in succession, designated respectively by mki, mi <2, ..., mj ,. .., mK.

According to the invention, the pronunciation speed of the words of the sentence PFlu is evaluated, that is to say the time required for the pronunciation of K words.

An identifier IVu representative of the value of this speed is then associated, in the subtitling component CSio, with the sentence PFlu, so that when they are returned to the screen, the K words of the sentence PHu appear then disappear each in turn at the speed associated with the identifier IVu, between the instant tu and the moment tu + i, respectively at K successive instants tki, tk2, ..., tK.

According to a first example, the indicator IVu is itself associated with a single appearance-disappearing time of a word which is the same for the K words.

According to a second example, the indicator IVu is associated with a time of appearance-disappearance of a word that is modulated according to the length of the word considered in the sentence, for example as a function of the number of syllables contained in the word . The subtitling step ST10 of FIG. 5 is thus repeated for each time interval [t1, t2], [t2, ta], and so on until the last time interval [tM, Tj.

According to a first variant embodiment of the second embodiment which is represented in FIG. 7A, in the case where the sentence PHu is not pronounced at constant speed, the sentence PHu is decomposed into as many sub-sentences as there has different speeds of pronunciation.

In the example of FIG. 7A, the sentence PHu is for example broken down into: a first subphrase PHui containing P words mki, mk2, ···, mkP, to which is associated a first identifier IVi representative of the speed of pronunciation of the words of the first subphrase PHui, - and a second sub-sentence PHu2 containing KP words mkP + i, mkp + 2, ···, ηπκ, with which is associated a second identifier IVu2 representative of the speed of pronunciation of the words in the second subphrase PHu2-

According to a second variant embodiment of the second embodiment which is represented in FIG. 7B, in the case where the plurality of K words consists of a first plurality of Q words mqi, mq2, ..., ma pronounced by a first speaker, at a given first speed Vi, and at least a second plurality of R words mr, mr2, ..., mp uttered by a second speaker, following the first speaker, at a given second speed V2, a first identifier IVi representative of the first pronunciation speed of the Q words is associated, in the subtitling component CS10, with the first plurality of Q words mqi, mq2, ..., ma, a second representative identifier IVu2 the second pronunciation speed of the R words is associated, in the subtitling component CS10, with the second plurality of R words mri, mr2, ..., mR.

Of course, in the case where the first plurality of Q words or the second plurality of R words is not pronounced at constant speed, the first plurality of Q words or the second plurality of R words is in turn decomposed into as many words. sub-sentences that there are different speeds of pronunciation, as in the first variant of the second embodiment, shown in Figure 7A.

According to a third variant embodiment of the second embodiment which is shown in FIG. 7C, similarly to the embodiment shown in FIG. 4B, in view of the fact that the same speaker utters the sentences PHui and PHu2 in a row. a lki identifier representative of the same speaker is inserted, in the subtitling component CS10, in association with at least the first word mki among said K words.

According to a fourth variant of the second embodiment which is represented in FIG. 7D, similarly to the embodiment shown in FIG. 4C, given that a first speaker pronounces the phrase PHui and then a second speaker pronounces the sentence PHu2: - a first identifier lqi representative of the first speaker is inserted, in the subtitling component CS10, in association with at least the first word mqi among said Q words of the sentence PHui, - a second identifier Im representative of the second speaker is inserted, in the subtitle component CS10, in association with at least the first word mn among said R words of the sentence PHu2. At the end of the subtitling step ST10 of FIG. 5, the stream F'i comprising the video component CV10, the audio component CA10, the subtitling component CS10, in association with the sub-title, is generated. SF'i flow. The flow F'i contains the temporal synchronization indicators ti, ta, ..., tu, ···, Im, as well as the word word speed indicators, and possibly the indicators representative of the speakers who pronounce the words.

The stream F'1 is then either transmitted to the terminal TER or to the access terminal STB of FIG. 1, via the network RC, or stored on a medium for a subsequent transmission.

A third embodiment of a method for generating audiovisual streams according to the invention will now be described with reference to FIG.

This third embodiment differs mainly from the embodiments shown in FIG. 2 and in FIG. 5, in that it does not contain a subtitling step.

As will be described later in the description, the subtitling is implemented here upon reception of the audiovisual stream transmitted via the communication network RC of FIG. 1.

Similarly to the embodiments shown in FIG. 2 and in FIG. 5, the flux generation method according to the third embodiment comprises an EDioo editing step, during which a flux F "i is constructed by insertion: a video component CVioo which contains all the video data contributing to the restitution of the images of the audiovisual content C, at least one audio component CAioo which contains all the audio data contributing to the complete sound reproduction of the audiovisual content C.

The video components CVioo and audio CAioo are inserted in a synchronized manner so that at a current moment of restitution of the audiovisual content, the sound emitted corresponds to the image restored on the screen.

As represented in FIG. 9: the video component CV100 contains a sequence of M video scenes SVi, SV2,..., SVu,..., SVm which scroll over the duration T of the content C, respectively at different successive instants. ti, t2, ..., tu, ..., tM, the audio component CA100 contains a sequence of M sets of audio data Di, D2, ..., Du, ... Dm which succeed one another over the entire duration T of the content C and which are respectively synchronized with the M video scenes of the video component CV100, at the different successive instants ti, t2,..., tu,..., tM which constitute temporal indicators of synchronization between the data of the video component CV100 and the audio data of the audio component CA100.

Similarly to the embodiments shown in FIG. 2 and in FIG. 5, the flux generation method according to the third embodiment further comprises a signaling step SIG100 represented in FIG. 8, during which it is conventionally carried out. the insertion, in a signaling sub-flow SF "i associated with the flow F" i, of a plurality of information allowing the analysis of the flow F "i by the terminal TER or the access terminal STB of the user.

With reference to FIG. 9, in which the sub-stream SF "i is represented, as in the first and second embodiments, such information is in particular metadata, such as for example an identifier ID of the audiovisual content C, information of DESC description of the audiovisual content C, such as its genre (film, documentary, sport, etc ...), the name and surname of the people associated with the audiovisual content C (director, actors, actresses, sportsmen, etc ...) .), associated temporal information, such as, for example, the duration T of the audiovisual content C, the start and end time of the broadcast of the content C if the latter is broadcast in real time, etc.

According to the third embodiment, in order to compensate for the absence of a subtitling component in the audiovisual stream to be transmitted to the TER terminal or to the access terminal STB of FIG. 1, the signaling step SIG100 is completed by the inserting information relating to the speaker (s) involved in the content C. For this purpose, for a plurality of W speakers LOCi, LOC2, ..., LOCi,... LOCw determined in the content C: the speaker LOCi is associated with a representative information INFV1 the voice frequency of the latter, and at least one morphological characteristic CMi associated with said speaker LOC1, the speaker LOC2 is associated with an information INFV2 representative of the voice frequency of the latter , as well as at least one morphological characteristic CM2 associated with said speaker LOC2,

J - the speaker LOC, is associated with information INFV, representative of the voice frequency of the latter, and at least one morphological characteristic CM, associated with said LOC speaker,

... the speaker LOCw is associated with a representative information INFVw the voice frequency of the latter, and at least one morphological characteristic CMw associated with said speaker LOCW-

Thus, as shown in FIG. 9, are inserted into the sub-stream SF "i: - a header LOC1, followed by the data fields INFV1 and CMi, - a header LOC2, followed by the data fields INFV2 and CM2

J - a LOC header, followed by data fields INFV, and CM ,,

J - a LOCw header, followed by the data fields INFVw and CMw-A speaker can for example be: - a person visible in the current video scene (eg actor or an actress in the case for example of a film) - a person out of the field (ex: voice "off" in the case for example of an animal documentary, a tennis match, etc ...), - etc ...

In the case where a speaker considered LOC, appears in the current video scene at the moment when he pronounces a word, the morphological characteristic CM, consists of image data of a part of the body of the speaker, for example the face of the last.

In the case where a speaker considered LOC does not appear in the current image at the moment when he pronounces a word, the data field CM is empty. The speaker LOC, is then associated with an identifier ZI, which is representative of the coordinates of a current video area capable of displaying each word spoken by the speaker LOC. Such coordinates may for example be those of a rectangle disposed at the bottom and in the center of the current image restored. At the end of the signaling step SIG100 of FIG. 8, the flux F "i comprising the video component CV100, the audio component CA100, in association with the sub-flux SF" i, is then generated. The stream F "i also contains the time synchronization indicators ti, ta, ..., tu, ..., tM-

The stream F "1 is then either transmitted to the terminal TER or to the access terminal STB of FIG. 1, via the network RC, or stored on a medium for a subsequent transmission.

According to a first variant embodiment, for a speaker considered LOCi: - the data field INFV, directly contains the value of the voice frequency of said speaker, - the data field CM, directly contains the image data of a part of the body of said speaker.

According to a second variant embodiment, for a speaker considered LOC: - the data field INFV, contains an access address, for example a URL link (of the "Uniterm Resource Locator"), to a database BDV as represented in FIG. 1, in which the voice print of said speaker has been stored beforehand in association with the identifier LOC, the data field CM contains an access address, for example a URL link, to a database BDI as shown in Figure 1, wherein were previously stored the image data of said speaker in association with the identifier LOCi.

In the example of FIG. 1, the databases BDV and BDI are both contained in a SER server connected to the communication network RC. As a variant, the databases BDV and BDI could be respectively contained in two separate servers connected to the communication network RC.

According to yet another alternative embodiment, the BDV and BDI databases could both be contained in the PFS service platform.

In relation to FIG. 10, we now consider the simplified structure of a REC device for receiving audiovisual content, such a device being contained in the content-rest terminal TER of FIG. 1, according to an exemplary embodiment of FIG. invention. Such a device for receiving audiovisual content is adapted to implement the audiovisual content reception method according to the invention which will be described below.

For example, the device REC comprises physical and / or software resources, in particular a processing circuit CT for implementing the method for receiving audiovisual content according to the invention, the processing circuit CT containing a processor PROC driven by a PG computer program. At initialization, the code instructions of the computer program PG are for example loaded into a RAM, denoted MR, before being executed by the processing circuit CT.

According to the invention, for an audiovisual content C to be restored by the terminal TER, the processing circuit CT is arranged to implement: the reception, via a reception interface RCV, of an audiovisual stream Fi or F ' i or F "i corresponding to the content C to be restored, such as for example transmitted by the PFS platform, - the analysis of the received stream, via an ANA analyzer, - a CAL calculator to calculate the moment of restitution, in the content C, of each word to be restored, - the generation of a subtitling component, in the case where the audiovisual stream received does not contain any, via a GCS generator of subtitling component, - the sending to TER terminal, via a COM command interface, a C content return command associated with the stream analyzed.

In relation to FIG. 11, the steps of the method for receiving audiovisual content according to a first embodiment according to the invention are now presented.

According to this first embodiment shown, said reception method is implemented by the device REC of FIG. 10.

During a step E1 shown in FIG. 11, the RCV interface of FIG. 10 receives the stream Fi as generated according to the method of FIG. 2, then transmitted via the communication network RC of FIG. Fi stream corresponding to a content C to be returned by a terminal TER of the user UT, such as for example a tablet.

During a step E2 shown in FIG. 11, the analyzer ANA of FIG. 10 reads in the sub-flow SFi associated with the stream Fi the information representative of the audiovisual content C that has been inserted during the signaling step SIG1 of FIG. 2. Such information is in particular metadata, such as, for example, an identifier ID of the audiovisual content C and description information DESC.

During a step E3 shown in FIG. 11, the ANA analyzer of FIG. 10 analyzes, one after the other, the time intervals [t 1, t 2], [t 2, t 1], and so on until at the last time interval [tM, T], in each of which a video scene of the video component CVi is synchronized with: - one or more corresponding delivered audio data inserted into the associated audio component CAi, - one or more words retranscribing the or the corresponding delivered audio data.

According to the first embodiment of the reception method, for a current time interval [tu, tu + i] analyzed, during which a current video scene SVu appears, and in the case where the corresponding set of audio data D the audio component CAi contains audio data representative of a plurality of K words mki, mk2, ..., mj, ..., mK (K ^ 2) constituting one or more sentences in the course of a step E4 shown in FIG. 11, the analyzer ANA successively identifies K words indexed respectively at K instants tki, tk2, ···, tK-. The step E4 is thus repeated for each time interval considered [t1, t2], [t2, ts], and so on until the last time interval [tM, T].

During a step E5 shown in FIG. 11, at each of the aforementioned K instants, a textual reproduction command of the corresponding indexed word is sent to the terminal TER, by means of the communication interface COM of FIG. 10.

Thus, the above-mentioned K words are advantageously reproduced textually one by one on the screen of the terminal TER, between two successive instants tu and tu + 1 of appearance of the current video scene SVu, a word restored at a given instant among the K instants replacing the word restored at the instant immediately preceding said moment considered. Step E5 is repeated for each time interval [t1, t2], [t2, ts], and so on until the last time interval [tM, T].

In this way, the viewing comfort of the user is significantly improved compared to the captioning of the prior art which proposes the simultaneous display of K words in the current time interval [tu, tu + i]. Indeed, thanks to the invention, instead of having to look constantly from left to right to read, in the current time interval [tu, tu + i], the subtitle sentences, l The user sees and reads only one word at a time.

Referring to FIG. 12A, there is shown a first example of reproduction, between two successive instants tu and tu + i, of a video scene SVu subtitled.

In the illustrated example, twelve words are uttered subsequently in the current time interval [tu, tu + i], by the actress present in the video scene SVu. These twelve words constitute for example the following sentence: "Make the husband and you dive so quickly that you will never go back! >>.

According to the invention: - tki is reproduced textually the word mki "Do >>, - tk2 is reproduced textually the word mk2" le "which replaces the word" Faire ", - tM is reproduced textually the word mK" never ! "Which replaces the word" remonteras ".

The twelve words are for example restored successively at the bottom of the corresponding image and in the center of the latter.

According to a first variant of this first embodiment, in the case where the stream Fi has been generated as represented in FIG. 4A, during the step E4 of FIG. 11, the analyzer ANA further identifies, in the subtitle component CSi, K identifiers lki, Ik2, ···, Ik associated respectively with K words mki, mk2, ..., mK.

As already explained in relation with FIG. 4A, the K identifiers are such that: the identifier lki is representative of the speaker who pronounces the word mki, the identifier Ik2 is representative of the speaker who pronounces the word mk2. the identifier Ικ is representative of the speaker who pronounces the word rriK.

During the step E5, the command of textual restitution of a word which is sent to the terminal TER then contains the identifier of the speaker who pronounces said word, so that the word is displayed in visual correspondence with the speaker .

In the case where the speaker is not visible on the screen, the word is displayed in a neutral zone of the image displayed on the TER terminal screen, for example at the bottom of the image.

Referring to FIG. 12B, there is shown an example of reproduction, between two successive instants tu and tu + i, of a current video scene SVu subtitled, according to this first variant.

This example is the same as that of Figure 12A, except that each of the twelve words spoken is reproduced verbatim on the screen in visual correspondence with the actress present in the image displayed on the screen.

Such an arrangement allows the user to distinguish on the screen the speaker who pronounces the words, here the actress, not the actor.

According to a second variant of this first embodiment, in the case where the stream Fi has been generated as shown in FIG. 4B, during the step E4 of FIG. 11, the ANA analyzer of FIG. the sub-titling component CSi: at the current time tki, a dentifier lki associated with at least the first word mki, at the following current instants tk2 ···, tk, respectively the words mk2, ..., mK devoid of corresponding identifiers.

During the step E5, the textual reproduction command of the first word mki which is sent to the terminal TER at time tki then contains the identifier lki of the speaker who pronounces said word, so that the word is displayed. in visual correspondence with the speaker.

Then, at the following current instants tk2 ···, tK, the textual reproduction command of the words mk2,..., MK which is sent to the terminal TER does not contain an identifier. These words will then be reproduced textually in visual correspondence with the speaker associated with the identifier Iki identified in step E4.

According to a third variant of this first embodiment, the flow Fi which has been generated is that represented in FIG. 4C. For this purpose, between two successive current instants tu and tu + i in the subtitling component CSi, the plurality of K words has been decomposed into a first plurality of J words mji, mj2, ..., mj pronounced by a first speaker respectively at successive instants tji, tj2, ..., tj, and at least a second plurality of L words mn, mi2, ..., mL pronounced by a second speaker, following the first speaker, respectively at The successive moments tu, ti2, ..., k-

At least the first word mji among said J words is associated with an identifier lji of the first speaker who pronounces the J words.

At least the first word mn among said L words is associated with an identifier In of the second speaker who pronounces the L words.

Therefore, during the step E4 of FIG. 11, the analyzer ANA identifies in the subtitling component CS 1: at the current instant t i, the identifier j i of the first speaker associated with at least the first mji word, - at the following current instants tj2 ..., tj, respectively the words mj2, ..., mj devoid of corresponding identifiers, - at the current moment tu, the identifier In of the second speaker associated at least with the first word mn, - at the following current instants ti2 ..., k, respectively the words mi2, ..., mL devoid of corresponding identifiers.

During the step E5, the textual reproduction command of the first word mji which is sent to the terminal TER at the time tji then contains the identifier lji of the first speaker who pronounces said word, so that the word is displayed in visual correspondence with this first speaker.

Then, at the following current instants tj2 ..., tj, the textual reproduction command of the words mj2,..., Mj which is sent to the terminal TER does not contain an identifier. These words will then be textually reproduced in visual correspondence with the first speaker associated with the identifier lji identified in step E4.

Then, during the step E5, the textual reproduction command of the first word mn that is sent to the terminal TER at the instant you then contains the identifier In of the second speaker who pronounces said word, so that the word is displayed in visual correspondence with this second speaker.

Then, at the following current instants ti2 ..., ti, the text reproduction command of the words mi2,..., Mi which is sent to the terminal TER does not contain an identifier. These words will then be reproduced textually in visual correspondence with the second speaker associated with the identifier In identified in step E4.

Such a provision allows the user to distinguish on the screen which speaker speaks the word, when two or more people are present on the screen.

A second embodiment of a method for receiving audiovisual content according to the invention will now be described with reference to FIG.

According to this second embodiment shown, said reception method is implemented by the device REC of FIG. 10.

Similarly to the first embodiment shown in FIG. 11, the audiovisual content reception method according to the second embodiment comprises a step E10 represented in FIG. 13, during which the RCV interface of FIG. 10 receives the stream F'i as generated according to the method of FIG. 5, then transmitted via the communication network RC of FIG. 1, said stream F'i corresponding to a content C to be restored by a TER terminal of the user UT, such than for example a tablet.

Similarly to the first embodiment shown in FIG. 11, the method for receiving audiovisual content according to the second embodiment comprises a step E20 represented in FIG. 13, during which the ANA analyzer reads in the sub-stream SF. associated with the stream F'i the information representative of the audiovisual content G that was inserted during the signaling step SIG10 of FIG. 5. Such information is in particular metadata, such as for example an identifier ID of the audiovisual content. G and DESG description information.

Similarly to the first embodiment shown in FIG. 11, the audiovisual content reception method according to the second embodiment comprises a step E30 represented in FIG. 13, during which the ANA analyzer analyzes one after the other time intervals [t1, t2], [t2, ta], and so on until the last time interval [tM, T], in each of which a video scene of the video component GVio is synchronized with: - one or more corresponding delivered audio data inserted into the associated audio component GAio, - one or more words retranscribing the corresponding audio output data.

As in the first embodiment of the reception method, for a current time interval [tu, tu + i] analyzed, during which a current video scene SVu appears, and in the case where the corresponding set of audio data of the audio component GAio contains audio data representative of a plurality of K words mki, mk2, ..., mj, ..., mK (K ^ 2) constituting one or more sentences, in the course of a step E40 shown in Figure 13, the analyzer ANA successively identifies K words.

In the example shown, the set of K words constitutes a current sentence PHu pronounced between the moments tu and tu + i.

According to the second embodiment, during the step E40, the analyzer ANA further identifies, in the subtitling component GSio, an identifier IVu representative of the value of the pronunciation speed of the K words. Step E40 is thus repeated for each time interval [t1, t2], [t2, ta], and so on until the last time interval [tM, Tj.

During a step E50 shown in FIG. 13, a calculator CAL of the reception device REC of FIG. 10 calculates the K corresponding instants of appearance of the K words as a function of the value of the speed associated with the identifier IVu. identified in step E40.

During a step E60 shown in FIG. 13, at each of the K instants calculated, a textual reproduction command of the corresponding identified word is sent to the terminal TER, by means of the communication interface COM of FIG. word being restored at the speed whose value is associated with the identifier IVu, at the instant of appearance of the identified word which was calculated in step E50.

Thus, thanks to the presence, in the stream F'i, of the speed identifier IVu associated with the sentence PHu, the subtitling component CSio used can retain the same structure as the subtitling components of the prior art, while allowing, between two successive instants tu and tu + i of appearance of a current video scene SVu, a textual restitution of the aforementioned K words, one by one, at the same rate as the actual rhythm of pronunciation of said K words, a word restored at a given moment replacing the word restored at the instant immediately preceding said instant considered. Step E60 is repeated for each time interval [ti, tz], [t2, ta], and so on until the last time interval [tM, T].

Referring to FIG. 14A, there is shown a first rendition example, between two successive instants tu and tu + i, of a current video scene SVu subtitled, according to the second embodiment of the method for receiving audiovisual content according to the invention.

The current video scene SVu is the same as that shown in Figure 12A. The words pronounced between the moments tu and tu + i are also the same as those of figure 12A.

In the exemplary rendering of FIG. 14A, the identifier IVu is associated with a constant pronunciation rate value that is the same for the K words. As a result, regardless of the length of a word, all words are textually rendered at the same speed and the time intervals [tki, tka], [tk2, tk3]> ·. [tK-1, ti <] appearance-disappearance of a word are all equal.

In the exemplary rendering of FIG. 14B, the identifier IVu is associated with a pronunciation speed value of a word that is modulated according to the length of the word considered in the sentence, for example as a function of the number of syllables. contained in the word.

Thus, in FIG. 14B, the appearance-disappearing time for example of the third word mks "mariole", which contains three syllables, is longer than the appearance-disappearing time, for example, of the first word mki "Make", which contains a single syllable.

According to a first variant of the second embodiment, the flow F'i which has been generated is that represented in FIG. 7A. For this purpose, the sentence PHu pronounced between the instants tu and tu + i is for example decomposed into: a first subphrase PHui containing P words mki, mk2 ,. ., mkp, with which is associated a first identifier IVui representative of the pronunciation speed of the words of the first subphrase PHui, - and a second sub-sentence PHu2 containing KP words mkP + i, mkp + 2, · · · , mK, with which is associated a second identifier IVu2 representative of the pronunciation speed of the words of the second subphrase PHu2 ·

Therefore, during the step E40 of FIG. 13, the analyzer ANA further identifies in the subtitling component CS10: the first identifier IV that is representative of the value of the pronunciation speed of the P words of the first subphrase PHui, - the second identifier IVu2 representative of the value of the pronunciation speed of the KP words of the second subphrase PHu2-

Then during step E50 of FIG. 13: the respective P appearance instants of the P words are computed as a function of the value of the speed associated with the first identifier IVi identified in step E40, the Κ- Ρ respective onset instants KP words are calculated based on the value of the speed associated with the second identifier IVu2 identified in step E40.

Then during the step E60 of FIG. 13, a textual restitution command of each of the P words of the first subphrase PHui, at the speed corresponding to the identifier IVui, is sent to the terminal TER, by means of the communication interface COM of FIG. 10. Then a textual restitution command of each of the words KP words of the second sub-sentence PHu2, at the speed corresponding to the identifier IVu2, is sent to the terminal TER, by means of the COM communication interface of FIG.

According to a second variant of the second embodiment, the flow F'i which has been generated is that represented in FIG. 7B. For this purpose, the plurality of K words pronounced between the instants tu and tu + i is composed of a first plurality of Q words mqi, mq2, ..., ma pronounced by a first speaker, at a first given speed Vi, respectively at O successive instants tqi, tq2, ..., ta, and at least a second plurality of R words mr, mr2, ..., mp pronounced by a second speaker, following the first speaker, at a second given speed V2, respectively at R successive instants tri, tr2, ..., tp.

A first identifier IVi representative of the first pronunciation speed of the Q words is associated, in the subtitling component CS10, with the first plurality of Q words mqi, mq2, ..., ma.

A second identifier IVu2 representative of the second pronunciation speed of the R words is associated, in the subtitling component CS10, with the second plurality of R words mn, mr2, ..., mp.

Therefore, during the step E40 of FIG. 13, the analyzer ANA identifies in the subtitling component CS10: the first identifier IV that is representative of the value of the pronunciation speed of the Q words pronounced by the first speaker, - the second identifier IVu2 representative of the value of the speed of pronunciation of the R words uttered by the second speaker.

Then during the step E50 of FIG. 13: the Q instants of appearance of the Q words are calculated as a function of the value of the speed associated with the first identifier IVi identified in step E40, the R instants respective appearance of the R words are calculated as a function of the value of the speed associated with the second identifier IVu2 identified in step E40.

Then during the step E60 of FIG. 13, a textual reproduction command of each of the Q words, at the speed ^ / ^ corresponding to the identifier IVui, is sent to the terminal TER, by means of the interface of FIG. COM communication of FIG. 10. Then a textual restitution command of each word R words, at the speed V2 corresponding to the identifier IVu2, is sent to the terminal TER, by means of the communication interface COM of FIG. .

According to a third variant of the second embodiment, the flow F'i which has been generated is that represented in FIG. 7C. For this purpose, the sub-titling component CS10 of FIG. 7A furthermore contains a lki identifier representative of the speaker that pronounces the sentence PHu previously decomposed into subphrases PHui and PHu2. The identifier lki is associated at least with the first word. mki among the K words that make up the set of sentences PHui and PHu2-

Therefore, during the step E40 of FIG. 13, the analyzer ANA further identifies in the subtitling component CS10: the first identifier IV that is representative of the value of the pronunciation speed of the P words of the first subphrase PHui, - the identifier lki associated with the first word mki of the first subphrase PHui and representative of the single speaker who utters after the two subphyses PHui, then PHu2, - the second identifier IVu2 representative of the value of the pronunciation speed of the KP words of the second subphrase PHu2 ·

Then during the step E60 of FIG. 13, a textual restitution command of each of the P words of the first subphrase PHui, at the speed corresponding to the identifier IVui, is sent to the terminal TER, by means of the communication interface COM of FIG. 10. The command for textual reproduction of the first word mki, which is sent to the terminal TER at the speed corresponding to the identifier IVui, then contains the identifier Iri of said single speaker who utters the sentence PHu, so that the word is displayed in visual correspondence with this speaker, at a time tki calculated according to the value of the speed corresponding to the identifier IVui.

Then, in step E60, the textual reproduction control of each of the following words mk2, ..., mkP of the subphrase PHui which is sent to the terminal TER does not contain a speaker identifier. These words will then be reproduced textually in visual correspondence with the unique speaker associated with the identifier Iki identified in step E40, at the speed corresponding to the identifier IVi identified in step E40, respectively at times tk2, ·· ·, TkP which were calculated in step E50, as a function of the value of the speed corresponding to the identifier IVui.

Then, the textual reproduction command of each of the following words mkP + i, ..., mK of the subphrase PHu2 that is sent to the terminal TER does not contain a speaker ID either. These words will then be textually reproduced in visual correspondence with the unique speaker associated with the identifier lki identified in step E40, at the speed corresponding to the identifier IVu2 identified in step E40, respectively at the times tkp + i, ..., tK which were calculated in step E50, as a function of the value of the speed corresponding to the identifier IVu2-

According to a fourth variant of the second embodiment, the flow F'i which has been generated is that represented in FIG. 7D. For this purpose, the sub-titling component CSio of FIG. 7B furthermore contains: an identifier IVi representative of the first pronunciation speed of the first plurality of Q words mqi, mq2, ..., ma. the identifier lqi associated with the first word mqi of the first sentence PHui and representative of the first speaker that pronounces the sentence PHui; - an identifier IVu2 representative of the second pronunciation speed of the second plurality of R words mn, mr2, .. ., mp, - the identifier Im associated with the first word mn of the second sentence PHu2 and representative of the second speaker who pronounces the sentence PHu2 ·

Then during the step E60 of FIG. 13, a textual restitution command of each of the Q words of the first sentence PHui, at the speed corresponding to the identifier IVui, is sent to the terminal TER, by means of the communication interface COM of FIG. 10. The command for textual reproduction of the first word mqi which is sent to the terminal TER, at the speed corresponding to the identifier IVui, then contains the identifier lqi of the first speaker that pronounces the sentence PHui, so that the word is displayed in visual correspondence with this first speaker, at a time tqi calculated according to the value of the speed corresponding to the identifier IVui.

Then, the textual reproduction command of each of the following words mq2, ..., mQ of the sentence PHui which is sent to the terminal TER does not contain a speaker identifier. These words will then be reproduced textually in visual correspondence with the first speaker associated with the identifier Iqi identified in step E40, at the speed corresponding to the identifier IVi identified in step E40, respectively at times tq2, ... , ta which were calculated in step E50 as a function of the value of the speed corresponding to the identifier IVui.

Then, the textual reproduction control of each of the following R words mri, ..., mp of the sentence PHu2, at the speed corresponding to the identifier IVu2, is sent to the terminal TER. The textual reproduction command of the first word mri which is sent to the terminal TER, at the speed corresponding to the identifier IVu2, then contains the identifier Li of the second speaker who pronounces the sentence PHu2, so that the word is displayed. in visual correspondence with this second speaker, at a sorting instant calculated according to the value of the speed corresponding to the identifier IVu2.

Then, the textual reproduction command of each of the following words mr2, ..., mR of the sentence PHu2 which is sent to the terminal TER does not contain a speaker identifier. These words will then be reproduced textually in visual correspondence with the second speaker associated with the identifier Im identified in step E40, at the speed corresponding to the identifier IVu2 identified in step E40, respectively at the times tr2, ... , tp which were calculated in step E50 as a function of the value of the speed corresponding to the identifier IVu2 ·

A third embodiment of a method for receiving audiovisual content according to the invention will now be described with reference to FIG.

According to this third embodiment shown, said reception method is implemented by the device REC of FIG. 10.

During a step El 00 shown in FIG. 15, the RCV interface of FIG. 10 receives the stream F "i as generated according to the method of FIG. 8, then transmitted via the communication network RC of FIG. 1, said stream F "i corresponding to a content C to be returned by a terminal TER of the user UT, such as for example a tablet.

During a step E200 shown in FIG. 15, the analyzer ANA of FIG. 10 reads in the sub-stream SF "i associated with the stream F" i the information representative of the audiovisual content C that has been inserted during the SIG100 signaling step of Figure 8. In known manner, such information is in particular metadata, such as for example an IDi identifier of the audiovisual content C and DESCi description information.

In addition, according to the third embodiment of the invention, the analyzer ANA identifies in the sub-stream SF "i: - in association with the field LOCi of data relating to a first speaker involved in the content, the information INFVi representative of the voice frequency of the latter, at least one morphological characteristic CMi associated with said first speaker, and a corresponding image area identifier Z \ ^ (if first speaker does not appear on the screen), - in association with the field LOC2 of data relating to a second speaker intervening in the content, the information INFV2 representative of the voice frequency of the latter, at least one morphological characteristic CM2 associated with said second speaker, and a corresponding image area identifier ZI2 (if the second speaker does not appear on the screen),

J - in association with the LOC field, data relating to an i-th speaker involved in the content, INFVi information representative of the voice frequency of the latter, at least one morphological characteristic CMi associated with said i-th speaker, and a ZI identifier, corresponding image area (if the i-th speaker does not appear on the screen),

J - in association with the LOCw field of data relating to a W-th speaker involved in the content, the information INFVw representative of the voice frequency of the latter, at least one morphological characteristic CMw associated with said W-th speaker, and a Zlw identifier of corresponding image area (if W-th speaker does not appear on the screen).

During a step E300 shown in FIG. 15, at a current instant, the analyzer ANA of FIG. 10 identifies: in the audio component CS100 of the stream F 1, one or more audio data delivered at said current instant and representative a word uttered by a speaker intervening in the content, - in the video component CV100 of the stream F "i, the corresponding video scene, synchronized with the audio data or delivered to said current time.

During a step E400 shown in FIG. 15, at said current instant, the subtitle component GCS generator of FIG. 10 textually converts said audio data or data delivered to said current instant in a current word.

According to an exemplary embodiment, the GCS generator consists of a well-known voice recognition algorithm per se.

During a step E500 shown in FIG. 15, at said current instant, the analyzer ANA identifies the current speaker who pronounces this word, for example the i-th speaker, by reading in the sub-stream SF "i, audit instantaneous current, of the field LOC, of data relating to the ith speaker, in association with the information INFVi representative of the vocal frequency of the latter, the morphological characteristic CM, of said i-th speaker, and a zone identifier ZI, of corresponding picture (if the i-th speaker does not appear on the screen).

During a step E600 represented in FIG. 15, at said current instant, the reception device REC triggers the sending to the terminal TER, by means of the communication interface COM, of a synchronized reproduction control of the said audio data, the corresponding video scene and said word, the restitution of said word being a function of information INFVi, Ci, Zl | read in sub-stream SF "i.

The steps E300 to E600 are repeated throughout the duration T of the audiovisual content C.

In order to predict the situation where a user would access the content a few moments after the current instant, the information INFVi, Ci, Zli are regularly refreshed for a predetermined duration.

Thanks to this third embodiment, the subtitling component is generated simultaneously with the analysis of the audio components CA100 and video CV100, each word being rendered textually on the screen one after the other and in visual correspondence with the speaker who pronounces the current word, a word restored at a given moment replacing the word restored at the instant immediately preceding said given instant.

According to a first variant embodiment, during step E500, the analyzer ANA reads in the data field INFV, the value of the voice frequency of the i-th current speaker, and, in the data field CM ,, the image data of a body part of said speaker, e.g. image data of the face of the current speaker.

An image recognition is then implemented in the video scene associated with the audio data analyzed in step E300, in order to identify a video zone that is close to the face of the current speaker and in which to reproduce the word verbatim. current.

According to a second variant embodiment, during step E500, the analyzer ANA: reads in the data field INEV, an access address, for example a URL link, to the database BDV of FIG. 1 , in which the voice print of said i-th current speaker has previously been stored in association with the information LOC, reads in the data field CM an access address, for example a URL link, at the base BDI data of FIG. 1, in which the image data of said i-th speaker in association with the LOC information have previously been stored, such image data being for example data of a facial image of the i-th current speaker. The communication interface COM or any other appropriate communication interface (not shown) of the reception device REC of FIG. 10 then sends: a request to supply the voiceprint of the i-th current speaker to the database BDV, - a request to supply the image data of the i-th current speaker to the database BDI.

Once the voice print and the image data of the i-th current speaker have been received in the reception device REC in response to the sending of said requests, an image recognition is then implemented in the a video scene associated with the audio data analyzed in step E300, in order to identify a video zone that is close to the face of the current speaker i-th and in which to restore the current word verbatim.

It goes without saying that the embodiments which have been described above have been given for purely indicative and non-limiting purposes, and that many modifications can easily be made by those skilled in the art without departing from the scope. of the invention.

Claims (10)

A method for receiving audiovisual content from an audiovisual stream that includes a video component and an audio component, a subtitle component being generated in association with the audio component and containing a sequence of words representative of the audio data of the audio component, said method being characterized in that it implements a restitution of the content, during which between two successive instants of restitution of a current video scene of the video component, in the case where the current audio data delivered between said two successive instants are representative of a plurality of K words (mki, mk2, .., mk), such that K> 2, said K words are restored between said two successive instants, respectively at K successive times, a word restored at a given instant replacing the word restored at a time immediately preceding said given instant. 2. A reception method according to claim 1, wherein the subtitling component being generated prior to the return of the content, each of said K words, pronounced at said successive K instants (tki, tk2, .., tK), having been associated prior to an identifier representative of the speaker who pronounces said word, the restitution of said content at a time considered among the K instants, is implemented: - by a prior analysis of the identifier corresponding to said time considered, - by a trigger of a display command, at the instant considered, of the word associated with said analyzed identifier, in visual correspondence with the speaker associated with said analyzed identifier. 3. Reception method according to claim 1, in which the subtitling component is generated prior to the reproduction of the content and the plurality of K words being pronounced by the same speaker at said successive K instants (tki, tk2, ··, ΐκ), the first word of said plurality of K words having also been previously associated with an identifier representative of said same speaker, the restitution of said content is implemented: at the first instant among said K instants, by a prior analysis of the identifier corresponding to said first word of the plurality of K words, - by triggering a command for successive display of said K words, respectively at K instants, in visual correspondence with said same speaker associated with said analyzed identifier. 4. Reception method according to claim 1, wherein the subtitling component being generated prior to the return of the content and the plurality of K words consisting of: - a first plurality of J words (mji, mj2 ,. ., mj) pronounced respectively at J instants (tji, tj2, .., tj) by a first speaker, - by at least a second plurality of L words (mn, mi2, .., mO pronounced respectively at L moments ( tu, ti2, .., ti) by a second speaker, following the J words uttered by the first speaker, the first word of said plurality of J words having previously been associated with a first identifier representative of the first speaker and the first word of said plurality of L words having been previously associated with a second identifier representative of the second speaker, the restitution of said content is implemented: - at the first moment of the plurality of J instants, by a prior analysis of the first identifier ass associated with said first word of the plurality of J words, - by triggering a command for successive display of said J words, respectively at J instants, in visual correspondence with the first speaker associated with said first identifier analyzed, - at the first instant of the plurality of L instants, by a preliminary analysis of the second identifier associated with said first word of the plurality of L words, - by triggering a command for successive display of said L words, respectively at L moments, in visual correspondence with the second speaker associated with the second identifier analyzed. 5. Reception method according to claim 1 or claim 2, wherein the subtitling component being generated prior to the return of the content and the plurality of K words being pronounced by the same speaker at said successive K instants (tki, tk2). , ··, k), at a given speed, said speed having also been associated in advance with an identifier representative of the value of said given speed, the restitution of said content is implemented: by a prior analysis of the identifier representative of the value of said given speed, - by a calculation, as a function of the value of the speed associated with the analyzed identifier, of each of the K instruction instants of the K words, - by a triggering of a command of successive display of said K words, respectively at K instants calculated. 6. Reception method according to claim 1, in which the subtitling component is generated simultaneously with an analysis of the content by implementing the following: identification, at a given moment among the K instants of pronunciation of the K words: • of one or more audio data of the audio component delivered at said instant and representative of one of the K words, • of the corresponding video scene synchronized with the one or more audio data, - textual conversion of the audio data delivered at said instant in one word. triggering a synchronized reproduction control of the one or more audio data, of the corresponding video scene and of said word, in association with said instant in question. 7. Reception method according to claim 6, in which the analysis of the content comprises, at the instant considered among the K instants of pronunciation of K words: an identification of the vocal frequency associated with the corresponding audio data or data delivered and representative of one of said K words uttered by a speaker considered, - a matching of said identified voice frequency with information relating to said speaker concerned which have been associated prior to said voice frequency, said reception method implementing, auditing moment considered, a triggering of a display command of said word corresponding to the audio data / delivered, in visual correspondence with the speaker considered. A device (REC) for receiving audiovisual content from an audiovisual stream that includes a video component and an audio component, a subtitle component being generated in association with the audio component and containing a series of words representative of audio data of the audio component, the device being characterized in that it comprises a processor which is arranged to implement a restitution of the content, during which between two successive instants of restitution of a current video scene of the component video, in the case where the current audio data delivered between said two successive instants are representative of a plurality of K words (mki, mk2, .., mK), such that K> 2, said K words are restored between said two successive instants, respectively at K successive instants, a word restored at a given instant replacing the restored word at a time immediately preceding said i nstant given. A computer program comprising program code instructions for performing the steps of the audiovisual content receiving method according to any one of claims 1 to 7, when said program is executed on a computer. A computer-readable recording medium on which a computer program is recorded including program code instructions for performing the steps of the audiovisual content receiving method according to any one of claims 1 to 7, when said program is executed by a computer.